The rapid adoption of connected vehicle technologies and advanced driver assistance systems (ADAS) necessitates robust security mechanisms capable of identifying and mitigating sophisticated cyber threats in real-time. Traditional signature-based intrusion detection systems (IDS) are often inadequate in addressing the dynamic and evolving nature of automotive cybersecurity threats, particularly in modern vehicle networks like Controller Area Network (CAN), CAN with Flexible Data-Rate (CAN-FD), and Automotive Ethernet. This research introduces a novel Real-time Intrusion Detection System utilizing advanced Machine Learning (ML) techniques designed specifically for automotive network environments.
The proposed IDS framework employs supervised and unsupervised ML algorithms, including anomaly detection, behavioral analytics, and predictive threat modeling, to achieve high accuracy and rapid threat identification capabilities. Through extensive testing in simulated and actual vehicle network scenarios, the developed IDS model demonstrates significant improvements over conventional detection methods, notably in precision, recall, detection latency, and adaptability to zero-day threats. This research further evaluates the proposed system's alignment with critical regulatory standards such as AIS 189 and UNECE WP.29, ensuring its practical applicability within automotive industry cybersecurity compliance frameworks. The findings highlight the potential for ML-driven IDS solutions to substantially enhance automotive cybersecurity posture, providing OEMs and stakeholders with actionable insights for proactive threat management.
Keywords: Automotive Cybersecurity, Intrusion Detection System, Machine Learning, CAN-FD, Automotive Ethernet, Predictive Analytics, AIS 189, UNECE WP.29, Real-time Threat Detection.